Skip to main content

Cornell University


Mueller Group Meetings 

Mueller group meetings are on Tuesdays at 2pm in PSB 416. Please see the Slack channel for further details.

General Computer Info

The new workstation is operational: It has the Intel C++ compiler, gcc,Anaconda, and Julia.

Useful links:

General Commands

To run a job so that it will keep running in background, and you can log out:

nohup myprogram >out.txt 2>err.txt &

One would type this instead of just myprogram

The nohup tells it not to "hang up" when you exit. The >out.txt redirects the output of the program into a text file "out.txt" and the 2>err.txt redirects the error messages from the program to a text file "err.txt". The & runs the program in the background.

If you forget to do the & just type control-Z (which suspends the job), and then bg, which puts it in the background. If you forget to do the nohup, you can type ps to get the process name (PID) of the running process, and then disown PID

To change the priority of a job type renice +19 -p PID The larger the number you use (+19) the lower the priority. It only goes up to 20, and starts at 1 so +19 puts it at lowest priority.

To see what is running type top, to see your processes type ps, To see what things you have explicitly ran, type jobs. ps and jobs do similar things, but ps gives you the process ID, while jobs gives you the jobs ID.

If you want to stop a job, you use kill PID. If that doesn?t work do kill -kill PID.

By default top shows priorities in the NI column.

For version control, I recommend checking out github,


Running Mathematica jobs from command line:

  1. In your notebook, highlight the cells you want run, and them to be "initialization cells" (control-8 or command 8 )
  2. Under "save as" choose ".m"
  3. Copy that .m file to mueller1 -- for example using scp
  4. use ssh to start a session on mueller1
  5. type nohup math whatever.m &

It will now run in the background. Make sure your script saves something to a file, because otherwise everything is lost. You can also redirect standard out and standard error.

Running Remote Jupyter Notebooks

To run a remote Jupyter notebook, you want to log-in to the server using port forwarding. For example: ssh -L 8881:localhost:8881 where you replace user with your user name. Then launch Jupyter with jupyter notebook --no-browser --port 8881. Then on your local machine launch a browser and navigate to http://localhost:8881/tree. You can change the 8881 to whatever is convenient.

Things to add

  • ssh public keys
  • Python: Conda environments Mathematica: Running mathematica scripts from command line (math or wolfram); connecting to a remote kernel
  • Compiled Languages: Using the compiler, linker, Makefiles?
  • Different Compilers: gcc, icc
  • CUDA
  • Git
  • curl, apt-get,
  • debuggers and profilers
  • IDEs